I am providing the detail setup of our replication.
We have defined 35 subscription set for processing 455 tables, grouped based on the business functionality.
Few tables have more than 1 Million records.
Our biggest table in size is 26 Million.
Problem appears with one table having 3.3 million, replication of this table seems to be in a loop.
This is were the Apply is initially performing insert and then swithing over to update. The update is strange. Maybe IBM SQL replication is internally designed to perform update when insert activity fails, this is my guess.
When we stop the subscription set containing this table, for all other subscription sets the data gets propagated. Now we have to troubleshoot with this information.
One thing I am surprised is that there are 10 to 20 tables which are bigger in size ( > 4 million) and they don't have this issue.
Why does replication fail for this table alone?